Markets + Stochastic

This example combines stochastic scenarios with multiple electricity markets, including stage-fixed decisions. It's the most complete example in the collection.

Source: examples/example5.py

What it demonstrates

• Stochastic scenarios with different market prices per scenario
• stage_fixed=True on the day-ahead market (same volumes in all scenarios)
• TradeDirection.BUY to restrict markets to buy-only
• How the optimizer splits decisions between "commit now" and "adjust later"

The setup

Two generators, a fixed load, and three buy-only markets:

• sdac: Day-ahead market, stage_fixed=True -- bids must be identical across scenarios
• sidc1: Intraday market 1 -- can adjust per scenario
• sidc2: Intraday market 2 -- can adjust per scenario

Two scenarios with different price forecasts.

Code

from datetime import timedelta

from odys.energy_system import EnergySystem
from odys.energy_system_models.assets.generator import PowerGenerator
from odys.energy_system_models.assets.load import Load, LoadType
from odys.energy_system_models.assets.portfolio import AssetPortfolio
from odys.energy_system_models.markets import EnergyMarket, TradeDirection
from odys.energy_system_models.scenarios import StochasticScenario

generator_1 = PowerGenerator(
    name="gen1",
    nominal_power=100.0,
    variable_cost=20.0,
    min_up_time=1,
    ramp_down=100,
)
generator_2 = PowerGenerator(
    name="gen2",
    nominal_power=150.0,
    variable_cost=100.0,
    min_up_time=4,
    min_power=30,
    startup_cost=0,
    ramp_up=140,
    ramp_down=100,
)

portfolio = AssetPortfolio()
portfolio.add_asset(generator_1)
portfolio.add_asset(generator_2)
portfolio.add_asset(Load(name="load", type=LoadType.Fixed))

energy_system = EnergySystem(
    portfolio=portfolio,
    markets=(
        EnergyMarket(
            name="sdac",
            max_trading_volume_per_step=150,
            stage_fixed=True,
            trade_direction=TradeDirection.BUY,
        ),
        EnergyMarket(
            name="sidc1",
            max_trading_volume_per_step=100,
            trade_direction=TradeDirection.BUY,
        ),
        EnergyMarket(
            name="sidc2",
            max_trading_volume_per_step=50,
            trade_direction=TradeDirection.BUY,
        ),
    ),
    scenarios=[
        StochasticScenario(
            name="scenario1",
            probability=0.5,
            available_capacity_profiles={
                "gen1": [100, 100, 100, 50, 50, 50, 50],
            },
            load_profiles={"load": [200, 75, 200, 50, 100, 120, 125]},
            market_prices={
                "sdac": [150, 150, 150, 150, 150, 150, 150],
                "sidc1": [200, 200, 200, 175, 100, 100, 100],
                "sidc2": [190, 150, 150, 175, 100, 100, 100],
            },
        ),
        StochasticScenario(
            name="scenario2",
            probability=0.5,
            available_capacity_profiles={
                "gen1": [100, 100, 100, 50, 50, 50, 50],
            },
            load_profiles={"load": [200, 75, 200, 50, 100, 120, 125]},
            market_prices={
                "sdac": [210, 210, 210, 185, 100, 100, 100],
                "sidc1": [120, 140, 140, 140, 140, 140, 140],
                "sidc2": [200, 160, 140, 180, 110, 90, 100],
            },
        ),
    ],
    timestep=timedelta(minutes=30),
    number_of_steps=7,
    power_unit="MW",
)

result = energy_system.optimize()

Reading the results

print(result.generators.power)     # dispatch per scenario
print(result.markets.sell_volume)   # (should be zero -- buy-only markets)
print(result.markets.buy_volume)    # volumes bought from each market
print(result.to_dataframe)

What to look for

• The sdac buy volumes are identical in both scenarios because stage_fixed=True. The optimizer picks volumes that work well on average.
• sidc1 and sidc2 volumes differ between scenarios since the optimizer can react to the realized prices.
• All markets are buy-only (TradeDirection.BUY), so sell_volume should be zero.
• Compare the sdac volumes across scenarios -- they're the same, which is the non-anticipativity constraint at work.